# Training-Free Model Ensemble for Single-Image Super-Resolution via Strong-Branch Compensation arXiv: https://arxiv.org/abs/2604.11564 · HTML: https://mlsanity.com/paper/2604.11564/ ## Key facts - Cited by 6 papers in the map (data through 14 Oct 2026). - Builds on 58 of its 76 references that are in the map. - First posted on arXiv 13 Apr 2026. ## Builds on (most cited first, 30 of 58) - [SwinIR: Image Restoration Using Swin Transformer](https://mlsanity.com/paper/2108.10257/) (2021) - [Restormer: Efficient Transformer for High-Resolution Image Restoration](https://mlsanity.com/paper/2111.09881/) (2021) - [Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data](https://mlsanity.com/paper/2107.10833/) (2021) - [VMamba: Visual State Space Model](https://mlsanity.com/paper/2401.10166/) (2024) - [Pre-Trained Image Processing Transformer](https://mlsanity.com/paper/2012.00364/) (2020) - [Uformer: A General U-Shaped Transformer for Image Restoration](https://mlsanity.com/paper/2106.03106/) (2021) - [Simple Baselines for Image Restoration](https://mlsanity.com/paper/2204.04676/) (2022) - [Designing a Practical Degradation Model for Deep Blind Image Super-Resolution](https://mlsanity.com/paper/2103.14006/) (2021) - [Exploiting Diffusion Prior for Real-World Image Super-Resolution](https://mlsanity.com/paper/2305.07015/) (2023) - [Activating More Pixels in Image Super-Resolution Transformer](https://mlsanity.com/paper/2205.04437/) (2022) - [SRDiff: Single Image Super-Resolution with Diffusion Probabilistic Models](https://mlsanity.com/paper/2104.14951/) (2021) - [MambaIR: A Simple Baseline for Image Restoration with State-Space Model](https://mlsanity.com/paper/2402.15648/) (2024) - [SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution](https://mlsanity.com/paper/2311.16518/) (2023) - [SinSR: Diffusion-Based Image Super-Resolution in a Single Step](https://mlsanity.com/paper/2311.14760/) (2023) - [Efficient Long-Range Attention Network for Image Super-resolution](https://mlsanity.com/paper/2203.06697/) (2022) - [Transformer for Single Image Super-Resolution](https://mlsanity.com/paper/2108.11084/) (2021) - [Dual Aggregation Transformer for Image Super-Resolution](https://mlsanity.com/paper/2308.03364/) (2023) - [Details or Artifacts: A Locally Discriminative Learning Approach to Realistic Image Super-Resolution](https://mlsanity.com/paper/2203.09195/) (2022) - [Efficient and Degradation-Adaptive Network for Real-World Image Super-Resolution](https://mlsanity.com/paper/2203.14216/) (2022) - [MambaIRv2: Attentive State Space Restoration](https://mlsanity.com/paper/2411.15269/) (2024) - [Blind Image Super-Resolution: A Survey and Beyond](https://mlsanity.com/paper/2107.03055/) (2021) - [SGS-SLAM: Semantic Gaussian Splatting For Neural Dense SLAM](https://mlsanity.com/paper/2402.03246/) (2024) - [Unsupervised Real-world Image Super Resolution via Domain-distance Aware Training](https://mlsanity.com/paper/2004.01178/) (2020) - [Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration](https://mlsanity.com/paper/2209.11345/) (2022) - [CiaoSR: Continuous Implicit Attention-in-Attention Network for Arbitrary-Scale Image Super-Resolution](https://mlsanity.com/paper/2212.04362/) (2022) - [Real-World Single Image Super-Resolution: A Brief Review](https://mlsanity.com/paper/2103.02368/) (2021) - [Omni Aggregation Networks for Lightweight Image Super-Resolution](https://mlsanity.com/paper/2304.10244/) (2023) - [Improving Image Restoration by Revisiting Global Information Aggregation](https://mlsanity.com/paper/2112.04491/) (2021) - [Spatially-Adaptive Feature Modulation for Efficient Image Super-Resolution](https://mlsanity.com/paper/2302.13800/) (2023) - [ResDiff: Combining CNN and Diffusion Model for Image Super-Resolution](https://mlsanity.com/paper/2303.08714/) (2023) ## Built on by (most cited first, 6 of 6) - [RAWild: Sensor-Agnostic RAW Object Detection via Physics-Guided Curve and Grid Modeling](https://mlsanity.com/paper/2605.05941/) (2026) - [FluxFlow: Conservative Flow-Matching for Astronomical Image Super-Resolution](https://mlsanity.com/paper/2605.03749/) (2026) - [ELoG-GS: Dual-Branch Gaussian Splatting with Luminance-Guided Enhancement for Extreme Low-light 3D Reconstruction](https://mlsanity.com/paper/2604.12592/) (2026) - [Dehaze-then-Splat: Generative Dehazing with Physics-Informed 3D Gaussian Splatting for Smoke-Free Novel View Synthesis](https://mlsanity.com/paper/2604.13589/) (2026) - [FujinSplat: Seeing Through Smoke with RAW-Domain Gaussian Splatting](https://mlsanity.com/paper/2609.06017/) (2026) - [ASTRA-SR: Atmospheric Seeing and Turbulence Restoration for Astronomical Image Super-Resolution](https://mlsanity.com/paper/2609.26731/) (2026) ## Abstract (arXiv) Single-image super-resolution has progressed from deep convolutional baselines to stronger Transformer and state-space architectures, yet the corresponding performance gains typically come with higher training cost, longer engineering iteration, and heavier deployment burden. In many practical settings, multiple pretrained models with partially complementary behaviors are already available, and the binding constraint is no longer architectural capacity but how effectively their outputs can be combined without additional training. Rather than pursuing further architectural redesign, this paper proposes a training-free output-level ensemble framework. A dual-branch pipeline is constructed in which a Hybrid attention network with TLC inference provides stable main reconstruction, while a MambaIRv2 branch with geometric self-ensemble supplies strong compensation for high-frequency detail recovery. The two branches process the same low-resolution input independently and are fused in the image space via a lightweight weighted combination, without updating any model parameters or introducing an additional trainable module. As our solution to the NTIRE 2026 Image Super-Resolution (× 4) Challenge, the proposed design consistently improves over the base branch and slightly exceeds the pure strong branch in PSNR at the best operating point under a unified DIV2K bicubic × 4 evaluation protocol. Ablation studies confirm that output-level compensation provides a low-overhead and practically accessible upgrade path for existing super-resolution systems.